SmolLM-135M-Instruct: benchmark results
Hugging Face's SmolLM-135M-Instruct, fine-tuned from the SmolLM-135M base model on public instruction datasets, part of the three-size SmolLM family trained on curated educational and synthetic data. Provider: HuggingFace. Released 2024-07-16. Access: Open.
Aggregate Rating 1150 ± 39, rank #1625 of 1656 rated models, from 36 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open CoT - LSAT Analytical Reasoning | 1.74 | CoT Gain (%) | 24.8 |
| Open LLM Leaderboard - MuSR | 36.35 | Score | 20 |
| Open CoT - LogiQA | 1.44 | CoT Gain (%) | 17.2 |
| Open CoT - LogiQA 2 | 1.53 | CoT Gain (%) | 15.3 |
| Open LLM Leaderboard - GPQA | 25.92 | Score | 11.8 |
| Open Portuguese LLM - ASSIN2 RTE | 39.45 | Macro F1 (%) | 11.8 |
| Open PL LLM - PolEmo2-OUT (generative, 5-shot) | 31.78 | Accuracy (%) | 10.4 |
| Open PL LLM - EQ-Bench | 0 | EQ-Bench Score | 9.7 |
| Open PL LLM - DYK (generative, 5-shot) | 18.39 | Binary F1 (%) | 9.4 |
| Open CoT - LSAT Reading Comprehension | 2.23 | CoT Gain (%) | 9.2 |
| Open PL LLM - PPC (generative, 5-shot) | 31.6 | Accuracy (%) | 9.1 |
| Open LLM Leaderboard - MMLU-Pro | 11.76 | Score | 8.8 |
Interactive version: theaggregate.ai/model?slug=smollm-135m-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.